devwithjev
β€” reading nowβ€” views
Submit a build

Jev Decision Layer for Agent Control

NO1ennn describes using Jev as a separate decision layer in an agent setup, with an orchestrator routing decisions through a toll booth to a fast lane or a human. Tool observations feed back into the flow.

View on X
NO1ennn@N01ennn𝕏
Jev in the control plane is the first agent setup I have run without sitting behind it 99% of people put an LLM on every decision and pay 400x for an answer that takes 3 seconds. 1% run the decision layer separately. the whole shape is one line: prompt β†’ orchestrator β†’ Jev decision + p β†’ toll booth β†’ fast lane or human β†’ tool β†’ observations back 8 steps and the architecture is done: step 1 β†’ log the action before you execute it, in the same DB transaction as the request. if the process dies, the action is still recorded. step 2 β†’ put it on kafka partitioned by session_id. one agent session sta
Sep 23, 2026X postsView on X
The author says to log an action before execution in the same database transaction as the request. They also describe putting actions on Kafka partitioned by session_id.

Also filed under Agents & browsers